Dynamical Downscaling over the Gulf of St. Lawrence using the Canadian Regional Climate Model

This study explores the problem of dynamical downscaling global climate model (GCM) simulations to finer resolution and compares the results with present climate reanalysis data. We use the Canadian Regional Climate Model (CRCM) to dynamically downscale outputs from the third generation Coupled Global Climate Model (CGCM3) for the Gulf of St. Lawrence and related coastal areas. The integration was performed for the 1970-99 tri-decadal period. Three methodologies were used: (DM-1) sea surface temperature (SST) fields and other fields from CGCM3 were used as drivers of the CRCM, (DM-2) CGCM3 SSTs were adjusted based on North American Regional Reanalysis (NARR) data to correct SST climate biases over the ocean, and (DM-3) CGCM3 SSTs were adjusted based on NARR data (as in (DM-2)) and longwave radiation was adjusted to remove biases in the CRCM's land surface temperature. The DM-1 methodology, using SSTs from CGCM3, gives surface temperature estimates that are too cold in summer and too warm in winter and underestimates 10 m winds in the summer. By comparison, the DM-2 and DM-3 methodologies produce more accurate estimates of marine winds and surface air temperature compared to the CGCM3 results, particularly in coastal areas. Differences among these methodologies are relatively minor at upper levels of the atmosphere.